[Avery Andrews 032698]
I've been doing something about my intention to learn more about
`Conversation Analysis' by going to a class in it that's being
given this semester at a manageable time, so here are some
thoughts/observation.
CA is an approach to the study of language developed by some
sociologists (Sacks, Schegloff & Jefferson are maybe the original
three big names) interested in interaction; one pillar of the
approach is to record natural interactions & study them in minute
details; experimentation is held to be unsound because the subjects
are presumed smart enough to be know they are being tested, and this
will warp the results (if you want to know what's going on in natural
conversation, what you have to look at is natural conversation, not
something different, such as talk-in-the-investigator's-laboratory).
Of course it's also unethical to record people without their knowledge;
the way around this is to make arrangements to the effect that there
will be some recording sometimes, but not say exactly when; then play
the actual recordings back to the subjects to make sure there's nothing
objectionable on them (I suspect privately that at least some serious
investigators make some secret recordings anyway, just to be sure!.)
Anyway, it seems to be the case that people forget that they're being
recorded pretty quickly, so that this doesn't have much of an effect on
the results>. Well there's the methodology. Now for some findings.
One is that in a wide range of conversational situations, there is an
extreme intolerance of periods of silence. If you meet with a colleage
in the hall, for example, and the conversation runs out of steam, you
quickly feel uncomfortable & either say someting more or come up with
a reason to break off the interaction. Likewise if the conversation
lags at the dinner table, people get pretty uncomfortable (in this
situation, you can't manufacture an excuse to bail). There are of
course other situations where constant talk isn't required (during sex,
or a long car-trip with friends); it would be useful to have a good
characterization of the ones where it is vs. the ones where it isn't.
I'll hand-wave at this issue by calling the situations requiring
continuous talk as `conversational situations'.
In a conversational situation, there seems to be a strong reference for
`perceiving somebody to be talking', it doesn't matter who: the rules
for ordinary conversation usually don't prescribe any particular person
to be the speaker (although there are local exceptions, as when somebody
asks somebody a question, then the ask-ee is supposed to respond). What
seems to happen is that when somebody finishes a `turn' (stops talking),
there may or may not be a short period of silence, but in
English-speaking culture, if it's longer than about a second, it gets
uncomfortable, and there seems to be a `problem' (`problem' seems to be
virtually a techninal term; I haven't figured out what it really means,
something to do with conflicting reference levels would be my guess).
The OK silence period varies between cultures; it is for example longer
in the country than in the city, which leads to city people thinking
that country people are stupid or sullen, since they don't answer
questions within the time expected by the city people, so that the
city folk think the country folk either don't know the answer or don't
want to provide it).
So it looks like there might be a `silence-perceiver' feeding into a
leaky integrator, & when the output of the integrator passes a certain
threshold, this triggers some sort of action.
Another relevant control system suppresses interruptions; the general
rule is that only one person is supposed to talk at a time. Something
that often happens after a silence is that several people start talking
at once, and then they all drop out except one, usually the one who
started first. This might be partially explained from the fact that
there's a bit of a lag between deciding to talk and having the first
words actually appear, but it remains to explain exactly why the
`don't interrupt' system has more of an effect on the people who
start in late than the one who starts first. Regardless of the details,
it seems to me that there might be something to discover about how the
control-systems of the individual participants interact to produce
the results.
Well, that's enough for starters.
Avery.Andrews@anu.edu.au